Deepfake technology, leveraging advanced machine learning and facial recognition algorithms, has significantly evolved since its popularisation in 2017 by a Reddit user. This innovation enables the creation of highly convincing fake videos and audio recordings, posing unprecedented challenges to information integrity, personal privacy, and democratic processes globally.
At its core, deepfake technology is utilised in two primary ways: "deepfaces" and "deepvoices." Deepfaces involve the replacement or creation of virtual faces in videos, while deepvoices alter or mimic voices. These advancements have unlocked new possibilities in media creation, offering tools for generating content that was once deemed impossible without the presence of the actual subjects. The technology has been particularly applied in the marketing domain, where it creates personalised and engaging content featuring celebrities like Taylor Swift, Selena Gomez, Elon Musk, and Joe Rogan, albeit without their direct participation.
Resurrecting Dead
Historically, the film and television industry has employed similar technologies for years, notably in resurrecting deceased actors for roles in movies such as Brian in "Fast and Furious 7" and Leia in "Rogue One." However, deepfake technology has evolved to not only replicate existing faces but also to create entirely fictional yet convincing ones. This evolution signifies the rapid advancements in AI's deep learning capabilities, which allow for the detailed study and recreation of human features, thereby blurring the lines between fiction and reality.
Exploits on Actors to International Actors
Noteworthy incidents of deepfake misuse include the dissemination of manipulated videos by pro-Russia social media accounts showcasing fabricated misconduct by Ukrainian soldiers. Similarly, viral deepfake videos of Indian actresses such as Rashmika Mandanna, Katrina Kaif, Kajol, and Alia Bhatt have raised concerns over artificial intelligence's potential for harm.
In the political arena, a deepfake video of Ukrainian President Volodymyr Zelenskyy in March 2022 misleadingly suggested a call for surrender, marking a high-profile instance of deepfake use in geopolitical misinformation.
Using Celebrities for Marketing
Despite its potential for innovation and creativity in marketing, the application of deepfake technology raises substantial ethical dilemmas. The primary concern revolves around the technology's ability to deceive audiences, spread misinformation, and undermine trust in media. The ethical considerations stem from the inherent risk of misusing these tools to manipulate public perception or to exploit the likeness of individuals without consent.
The innovative use of deepfake technology by consumer brands like Mondelez, ITC, and Zomato marks a significant shift in advertising strategies, allowing viewers to interact and co-star with prominent celebrities. This development showcases a blend of creativity and technology, opening new avenues for personalised and engaging advertisements.
Local Acts of Shahrukh Khan, Hrithik Roshan
Mondelez, in particular, achieved remarkable recognition at the Cannes Lions Festival of Creativity, securing multiple awards including India's first-ever Titanium Lion. The accolade was for its AI-generated Cadbury advertisement, which enabled local shop owners to feature alongside a deepfake of Shahrukh Khan. By submitting basic details about their business, these owners could receive a free personalised advertisement.
The AI utilised in this campaign crafted a deepfake of Shahrukh Khan's face and voice to specifically promote the local store, highlighting the potential of AI in creating tailored and meaningful advertising content. Similarly, Zomato leveraged deepfake technology to create an advertisement featuring Hrithik Roshan. In the ad, Roshan is seen expressing cravings for specific dishes from well-known restaurants across various cities. This campaign smartly integrated the viewers' phone GPS to identify top dishes and restaurants within their vicinity, delivering a customised viewing experience that directly linked the audience's location to the ad's content.
ITC also entered the deepfake advertising space through its collaboration with AI creative company Akool for the #HarDilKiFantasy campaign. This campaign was designed for ITC's Sunfeast Dark Fantasy biscuit brand and allowed participants to share the screen with Shahrukh Khan, further exemplifying the wide-ranging applications of deepfake technology in marketing.
Need for Regulation
The line between innovation and manipulation is thin, and navigating this landscape requires a careful consideration of the potential impacts on society, trust in media, and the respect for individual rights. The emergence of deepfake technology in marketing underscores the importance of developing guidelines and regulations that promote responsible use, ensuring that the advancements in AI serve to enhance, rather than undermine, the authenticity and integrity of digital content.
As brands continue to explore and expand the boundaries of advertising through deepfake technology, questions regarding legal protections and ethical considerations arise. Creating a deepfake ad involves navigating complex issues related to consent, copyright, and the potential for misinformation. It also underscores the need for comprehensive legal frameworks that safeguard the interests of individuals and brands while promoting innovation and creativity in advertising. This balance is crucial for ensuring that deepfake technology can be used responsibly, maintaining trust and authenticity in digital marketing efforts.
Government's Response
In India, with its 760 million internet users, a significant public outcry against fake videos has prompted governmental action. IT Minister Ashwini Vaishnav announced impending regulations to address the deepfake threat. The Ministry of Electronics and Information Technology's advisory on November 07, 2023, emphasises the need for social media intermediaries to identify and mitigate misinformation in accordance with the IT Rules 2021.
Societal Impact
The advancement of deepfake technology, facilitated by GANs (Generative Adversarial Networks) and machine learning, has democratised the creation of convincing deepfakes, enabling their spread with little to no coding knowledge. This ease of creation has profound implications for privacy, security, and the authenticity of digital content. Additionally, the misuse of deepfakes in financial scams, as reported by McAfee, indicates a significant risk of financial fraud involving AI voice cloning, report says about 47 per cent portion of the Indian adult population have experienced or know someone who have experienced some kind of AI voice scam.
EU's Regulatory Measures
In response to the escalating threats posed by AI, 29 countries, including the US, Canada, Australia, China, Germany, India, and members of the European Union, have pledged cooperation to mitigate potential harms through the Digital Personal Data Protection Act, 2023. This act emphasises the need for accountability among social media platforms in protecting sensitive personal data and ensuring the privacy of individuals. Researchers from Sungkyunkwan University in South Korea who studied the largest and most diverse dataset of real deepfakes in the world catalogued 2,000 deepfakes from 21 countries in English, Russian, Mandarin, and Korean languages. They said the deepfakes were created using the material sourced from Youtube, TikTok, Reddit, and Chinese video sharing platform Bilibili.
Challenges, Solutions
The rapid development of diffusion model deepfakes and generative AI presents new challenges, making deepfakes more realistic and accessible. Institutions like Stanford University and companies such as Microsoft have developed tools and educational resources to help detect and understand deepfakes, emphasizing the critical role of technology in combating misinformation.
Conclusion: Towards a Resilient Future
As the digital landscape continues to evolve, the proliferation of deepfakes calls for a multi-faceted approach, combining legal, technological, and educational efforts to safeguard democratic integrity and individual privacy. The collaborative research by experts from institutions like Sungkyunkwan University in South Korea highlights the importance of global efforts in understanding and countering the deepfake phenomenon. With the increasing sophistication of deepfakes, the imperative to maintain the authenticity of digital content and protect the foundational principles of democracy and security has never been more crucial.
